1. Field of the Invention
The present invention relates to a method of correcting a color image, the method being suitable for color correction (color proof) in a video printer, a digital color copying machine, or the like, and an apparatus for practicing the above method.
2. Description of the Prior Art
In order to obtain a hard copy of a television image signal by using a video printer, a digital color copying machine, or the like, a color image correction apparatus is often used to perform color correction so as to match the colors of the original image signal with the reproduced colors.
In a conventional color masking apparatus as a typical color image correction apparatus, a secondary absorbing component of a color material (e.g, a toner, an ink, a thermal transfer ink, and print paper) is canceled to reproduce accurate colors (intermediate colors).
In a television image, a color image is formed according to the additive primaries and the colors are represented according to a coordinate system of R, G, and B phosphors. However, a color image formed on print paper is reproduced according to the subtractive primaries. The colors are represented by a color system of L*, u* and v*. Therefore, signal data conversion (color correction) must be performed between these different color systems.
For example, in a color masking apparatus 1 shown in FIG. 1, R, G, and B color image data are arithmetically calculated to obtain new image data (the color-corrected image data, i.e., cyan (C), magenta (M), and yellow (Y) data). A color image is recorded on the basis of the new color data C, M, and Y.
Referring to FIG. 1, reference numeral 2 denotes a television receiver; 3, a color printer, and 4, a recording medium such as print paper.
A masking method used in such a color masking apparatus 1 is a linear masking method or a nonlinear masking method.
Linear masking employs a calculation represented by equation (1) and nonlinear masking employs a calculation represented by equation (2) as follows: ##EQU1## where A and D are coefficient matrices.
Linear masking is performed using a 3.times.3 matrix represented by equation (1). In order to realize this calculation, a multiplier is used to perform calculations step by step, or calculation results are formatted into a table (look-up table (LUT)) and color correction data is read out from the look-up table.
Nonlinear masking can be performed by using a logic array or the LUT described above.
The correction values obtained by the above-mentioned linear calculations are approximated values according to polynomial approximation, and the resultant color correction data is also inaccurate. In particular, in order to derive a coefficient A, key color matching is performed. This matching is relatively good for key colors but is not suitable for other colors, i.e., colors excluded from the key colors. As a result, the hue, the saturation, and the lightness are deviated from the correct values in the colors excluded from the key colors.
In particular, when image information from a television monitor is to be reproduced as a hard copy, large color reproduction errors occur even if key color matching is performed because the image signal from the television monitor is based on the additive primaries while the hard copy formed on the print paper is based on the subtractive primaries and color reproduction ranges between these color reproduction systems are different from each other.
When color correction data is obtained by nonlinear processing, the hue, saturation, and lightness errors of the colors excluded from the key colors are minimized, and good color reproducibility can be achieved.
However, in color correction according to nonlinear processing, complicated hardware is required. When an LUT or the like is used, a large memory capacity is undesirably required.
Assume that C, M, or Y image data is represented by 8-bit data. All combinations of these data are calculated as 2.sup.8.2.sup.8.2.sup.8 =2.sup.24. In order to obtain three color correction data, the required memory capacity is:
2.sup.24 .times.3=50.3 Mbytes
Japanese Unexamined Patent Publication (Kokai) No. 61-60068 discloses one means for solving the hardware problem and reducing the memory capacity.
An arrangement of this means is shown in FIG. 2. Input image data is divided into upper and lower bits. The upper and lower bit data are respectively supplied to corresponding LUT 11 and LUT 12 as reference address data. The upper bit LUT 11 stores a plurality of color correction data calculated in advance at proper intervals and serving as color correction data accessed by the upper bits of the input image data. The lower bit LUT 12 stores color correction data corresponding to a single correction curve.
The image data accessed by the upper bits of the input image data and the color correction data accessed by the lower bits of the input image data are added by an adder 13, thereby obtaining updated color correction data.
When the input image data is divided into the upper and lower bits and the upper and lower bits are subjected to processing, the required memory capacity can be greatly reduced as compared with the conventional memory.
However, the above means has the following disadvantage.
If output image data corresponding to the input image data is represented by a curve L.sub.1 shown in FIG. 3A and linear approximation color correction data (mark .smallcircle.) represented by a straight line L.sub.2 is used, an output error tends to be increased when the magnitude of the input value is increased.
If nonlinear approximation color correction data (mark .smallcircle.) having the same gradient as the input image data and represented by a curve L.sub.3 shown in FIG. 3B is used, the characteristic line (or curve) of the color correction data stored in the lower bit LUT 12 has a predetermined gradient (shape). Therefore, the data (output value) after color correction is represented by a curve L.sub.4.
Whether or not a portion of the input image data has a small or large gradient, the color correction data has a discrete portion and continuous color correction cannot be performed.
In a conventional color masking apparatus, a combination of basic colors (three or four colors) for reproducing a designated hue is obtained by accurately obtaining the hue characteristics of a color printer or the like. Therefore, the color conversion errors can be minimized and color reproduction characteristics can be greatly improved.
There are two conventional methods as a method of calculating a combination of basic colors (three or four colors) for reproducing a designated hue.
First, a density additive method is used. In order to form a hardcopy on print paper, spectral absorption densities of single colors (Y, M, and C) are measured, and total absorption characteristics are calculated using the measured densities according to the density additive method. The calculated total absorption characteristics are then converted into X, Y, and Z values, or L*, u*, and v* values in a color system. The density additive method is defined as a method of adding densities of the respective colors in spectral densities.
FIG. 4 is a graph showing special absorption density characteristics of colors. The special density versus wavelength is shown for two colors Y and M in solid line and solid/dashed line, respectively. The addend of the spectral density is shown in dashed line and designated Y and M;
Second, a Neugebauer equation is used to estimate a combination of the basic colors in printing.
The density additive method is not practiced in an actual system when print paper is used because it has poor precision of color reproducibility estimation.
Even if the Neugebauer equation is used, this produces an approximation and a difference between the approximated value and the actual value is large. Precision of color reproducibility is poor.
In order to solve the problem associated with precision of such color reproducibility, a method may be proposed to record a color image of a television image signal directly on a recording medium such as print paper, measuring the colors of the recorded image, and calculating the correction values.
In this case, measurement is preferably performed on the basis of color patch images, i.e., a plurality of color images obtained by combining a plurality of basic colors because the measured color points can be clearly specified.
However, the as number of color patches is indefinitely increased, the color measurement points are increased accordingly. The increase in color measurement points undesirably prolongs the color measurement time. The above solution is not the best solution.
If a color reproduction range of an output system such as print paper and a color reproduction range of an input system such as a color CRT are expressed by a single color system of L*, u*, and v*, the two color reproduction ranges differ from each other. More specifically, the color reproduction range of the output color system is narrower than that of the input color system.
FIG. 5 shows a relationship between a desired printed product and a practical print paper. The color reproduction range of the printed product is represented by a curve L.sub.1, and that of the print paper is represented by a curve L.sub.2.
If the lightness and saturation levels in the input color reproduction range are high and these levels are not included in the color reproduction range of the output system (region *1 in FIG. 5), the input colors cannot be expressed.
If color image information having a level exceeding the color reproduction range of the output color system is input, a corresponding value does not exist in the output color system. According to a conventional technique, color correction data associated with the color image information having a level exceeding the color reproduction range of the output color system is arbitrarily compressed within the color reproduction range. Therefore, the hue, the saturation and the lightness are changed in a direction different from that of the human visual sense. The output colors become unnatural.